Corollary 2 as a word RAM program. Read the word, sort the jobs by start time, sum the preprocessing times into $P$, and fill the dual table of Section 3 over sets of thresholds written as base-$(n+1)$ numbers: the numbers are visited from the top, each is decided to be a code of a set or not by a recurrence, and the two sets $X₁$, $X₂$ of recursion (1) are found by one scan of its digits ($O(|X|)$, independent of $P$), after which its $P+1$ cells cost $O(1)$ each. The table records, for each instant $t = 0 … P$, the largest weight attainable from $t$, capped at the threshold (so its entries never leave the word); the block of the empty set is zero, and the answer is the entry of the first $m$ indices at instant $0$. The sets contribute $Σ(|X|+1) ≤ 2 (n+1)^m$ in all. The domain restricts to positive processing times, the paper's standing assumption for Lemma 1's recursion. The running time is $c (P+1) (n+1)^m + c · sortCost$: no work per machine beyond the $O(|X|)$ scan of a set.